The effect of potassium removal by hemodialysis on the plasma potassium concentration (PK) was observed during 20 studies on 7 anuric patients. The following was observed: (1) The pattern of change in PK is consistent with at least a two-compartment distribution; hence, in contrast to urea, PK shows a marked postdialytic rebound and the simple equation for solute removal from a single pool cannot predict postdialysis PK. (2) The magnitude of the fall in PK correlated with the predialysis PK; this was not only due to increased potassium removal, but also due to a dependence of the apparent volume of distribution of potassium on the state of potassium balance. The fractional decrement in plasma potassium concentration, contrary to that of urea, is a function of predialysis concentration, and prediction of the effect of dialysis on plasma potassium concentration can assume neither a single pool nor a constant distribution volume for this ion.